Development of high accuracy congestion prediction algorithm using series of camera detectors

Abstract

This paper proposes a high accuracy algorithm to predict short-term traffic congestion in highway using patterns of microscopic traffic variable, including speed and its standard deviation from series of camera detectors installed at the location before observing point. The performance of the proposed algorithm is compared with a single camera detector. The result from simulation data shows the prediction accuracy of the proposed algorithm that utilizing data from series of detectors is twice that of single detector with zero false alarm rate. The proposed algorithm also performs well when applied to real-world data that show an increase of prediction accuracy by approximately fifty percent while achieve very low false alarm rate.